idnits 2.17.00 (12 Aug 2021) /tmp/idnits30205/draft-ietf-uta-rfc7525bis-05.txt: -(1130): Line appears to be too long, but this could be caused by non-ascii characters in UTF-8 encoding -(1345): Line appears to be too long, but this could be caused by non-ascii characters in UTF-8 encoding Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == There are 7 instances of lines with non-ascii characters in the document. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) -- The draft header indicates that this document obsoletes RFC7525, but the abstract doesn't seem to directly say this. It does mention RFC7525 though, so this could be OK. -- The draft header indicates that this document updates RFC6066, but the abstract doesn't seem to mention this, which it should. -- The draft header indicates that this document updates RFC5288, but the abstract doesn't seem to mention this, which it should. Miscellaneous warnings: ---------------------------------------------------------------------------- (Using the creation date from RFC5288, updated by this document, for RFC5378 checks: 2007-06-19) -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (3 February 2022) is 107 days in the past. Is this intentional? Checking references for intended status: Best Current Practice ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) -- Possible downref: Normative reference to a draft: ref. 'I-D.ietf-httpbis-semantics' == Outdated reference: draft-ietf-tls-dtls13 has been published as RFC 9147 ** Downref: Normative reference to an Informational RFC: RFC 4949 ** Obsolete normative reference: RFC 5246 (Obsoleted by RFC 8446) ** Obsolete normative reference: RFC 6347 (Obsoleted by RFC 9147) ** Downref: Normative reference to an Informational RFC: RFC 6979 ** Downref: Normative reference to an Informational RFC: RFC 7748 == Outdated reference: A later version (-14) exists of draft-ietf-tls-esni-13 == Outdated reference: A later version (-04) exists of draft-irtf-cfrg-aead-limits-03 -- Obsolete informational reference (is this intentional?): RFC 2246 (Obsoleted by RFC 4346) -- Obsolete informational reference (is this intentional?): RFC 4346 (Obsoleted by RFC 5246) -- Obsolete informational reference (is this intentional?): RFC 4347 (Obsoleted by RFC 6347) -- Obsolete informational reference (is this intentional?): RFC 5077 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 6961 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 7507 (Obsoleted by RFC 8996) Summary: 6 errors (**), 0 flaws (~~), 4 warnings (==), 12 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 UTA Working Group Y. Sheffer 3 Internet-Draft Intuit 4 Obsoletes: 7525 (if approved) P. Saint-Andre 5 Updates: 5288, 6066 (if approved) Mozilla 6 Intended status: Best Current Practice T. Fossati 7 Expires: 7 August 2022 arm 8 3 February 2022 10 Recommendations for Secure Use of Transport Layer Security (TLS) and 11 Datagram Transport Layer Security (DTLS) 12 draft-ietf-uta-rfc7525bis-05 14 Abstract 16 Transport Layer Security (TLS) and Datagram Transport Layer Security 17 (DTLS) are widely used to protect data exchanged over application 18 protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the 19 years, the industry has witnessed several serious attacks on TLS and 20 DTLS, including attacks on the most commonly used cipher suites and 21 their modes of operation. This document provides recommendations for 22 improving the security of deployed services that use TLS and DTLS. 23 The recommendations are applicable to the majority of use cases. 25 This document was published as RFC 7525 when the industry was in the 26 midst of its transition to TLS 1.2. Years later this transition is 27 largely complete and TLS 1.3 is widely available. Given the new 28 environment, updated guidance is needed. 30 Status of This Memo 32 This Internet-Draft is submitted in full conformance with the 33 provisions of BCP 78 and BCP 79. 35 Internet-Drafts are working documents of the Internet Engineering 36 Task Force (IETF). Note that other groups may also distribute 37 working documents as Internet-Drafts. The list of current Internet- 38 Drafts is at https://datatracker.ietf.org/drafts/current/. 40 Internet-Drafts are draft documents valid for a maximum of six months 41 and may be updated, replaced, or obsoleted by other documents at any 42 time. It is inappropriate to use Internet-Drafts as reference 43 material or to cite them other than as "work in progress." 45 This Internet-Draft will expire on 7 August 2022. 47 Copyright Notice 49 Copyright (c) 2022 IETF Trust and the persons identified as the 50 document authors. All rights reserved. 52 This document is subject to BCP 78 and the IETF Trust's Legal 53 Provisions Relating to IETF Documents (https://trustee.ietf.org/ 54 license-info) in effect on the date of publication of this document. 55 Please review these documents carefully, as they describe your rights 56 and restrictions with respect to this document. Code Components 57 extracted from this document must include Revised BSD License text as 58 described in Section 4.e of the Trust Legal Provisions and are 59 provided without warranty as described in the Revised BSD License. 61 Table of Contents 63 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 3 64 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 5 65 3. General Recommendations . . . . . . . . . . . . . . . . . . . 5 66 3.1. Protocol Versions . . . . . . . . . . . . . . . . . . . . 5 67 3.1.1. SSL/TLS Protocol Versions . . . . . . . . . . . . . . 5 68 3.1.2. DTLS Protocol Versions . . . . . . . . . . . . . . . 6 69 3.1.3. Fallback to Lower Versions . . . . . . . . . . . . . 7 70 3.2. Strict TLS . . . . . . . . . . . . . . . . . . . . . . . 7 71 3.3. Compression . . . . . . . . . . . . . . . . . . . . . . . 8 72 3.4. TLS Session Resumption . . . . . . . . . . . . . . . . . 8 73 3.5. Renegotiation in TLS 1.2 . . . . . . . . . . . . . . . . 9 74 3.6. Post-Handshake Authentication . . . . . . . . . . . . . . 10 75 3.7. Server Name Indication . . . . . . . . . . . . . . . . . 10 76 3.8. Application-Layer Protocol Negotiation . . . . . . . . . 11 77 3.9. Zero Round Trip Time (0-RTT) Data in TLS 1.3 . . . . . . 11 78 4. Recommendations: Cipher Suites . . . . . . . . . . . . . . . 12 79 4.1. General Guidelines . . . . . . . . . . . . . . . . . . . 12 80 4.2. Cipher Suites for TLS 1.2 . . . . . . . . . . . . . . . . 13 81 4.2.1. Implementation Details . . . . . . . . . . . . . . . 14 82 4.3. Cipher Suites for TLS 1.3 . . . . . . . . . . . . . . . . 15 83 4.4. Limits on Key Usage . . . . . . . . . . . . . . . . . . . 15 84 4.5. Public Key Length . . . . . . . . . . . . . . . . . . . . 16 85 4.6. Truncated HMAC . . . . . . . . . . . . . . . . . . . . . 17 86 5. Applicability Statement . . . . . . . . . . . . . . . . . . . 17 87 5.1. Security Services . . . . . . . . . . . . . . . . . . . . 18 88 5.2. Opportunistic Security . . . . . . . . . . . . . . . . . 19 89 6. Security Considerations . . . . . . . . . . . . . . . . . . . 19 90 6.1. Host Name Validation . . . . . . . . . . . . . . . . . . 19 91 6.2. AES-GCM . . . . . . . . . . . . . . . . . . . . . . . . . 20 92 6.2.1. Nonce Reuse in TLS 1.2 . . . . . . . . . . . . . . . 20 93 6.3. Forward Secrecy . . . . . . . . . . . . . . . . . . . . . 21 94 6.4. Diffie-Hellman Exponent Reuse . . . . . . . . . . . . . . 22 95 6.5. Certificate Revocation . . . . . . . . . . . . . . . . . 23 96 7. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 24 97 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 25 98 8.1. Normative References . . . . . . . . . . . . . . . . . . 25 99 8.2. Informative References . . . . . . . . . . . . . . . . . 27 100 Appendix A. Differences from RFC 7525 . . . . . . . . . . . . . 34 101 Appendix B. Document History . . . . . . . . . . . . . . . . . . 35 102 B.1. draft-ietf-uta-rfc7525bis-05 . . . . . . . . . . . . . . 36 103 B.2. draft-ietf-uta-rfc7525bis-04 . . . . . . . . . . . . . . 36 104 B.3. draft-ietf-uta-rfc7525bis-03 . . . . . . . . . . . . . . 36 105 B.4. draft-ietf-uta-rfc7525bis-02 . . . . . . . . . . . . . . 36 106 B.5. draft-ietf-uta-rfc7525bis-01 . . . . . . . . . . . . . . 36 107 B.6. draft-ietf-uta-rfc7525bis-00 . . . . . . . . . . . . . . 37 108 B.7. draft-sheffer-uta-rfc7525bis-00 . . . . . . . . . . . . . 37 109 B.8. draft-sheffer-uta-bcp195bis-00 . . . . . . . . . . . . . 37 110 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 37 112 1. Introduction 114 Transport Layer Security (TLS) and Datagram Transport Security Layer 115 (DTLS) are widely used to protect data exchanged over application 116 protocols such as HTTP, SMTP, IMAP, POP, SIP, and XMPP. Over the 117 years leading to 2015, the industry has witnessed serious attacks on 118 TLS and DTLS, including attacks on the most commonly used cipher 119 suites and their modes of operation. For instance, both the AES-CBC 120 [RFC3602] and RC4 [RFC7465] encryption algorithms, which together 121 were once the most widely deployed ciphers, have been attacked in the 122 context of TLS. A companion document [RFC7457] provides detailed 123 information about these attacks and will help the reader understand 124 the rationale behind the recommendations provided here. That 125 document has not been updated in concert with this one; instead, 126 newer attacks are described in this document, as are mitigations for 127 those attacks. 129 The TLS community reacted to these attacks in several ways: 131 * Detailed guidance was published on the use of TLS 1.2 [RFC5246] 132 and DTLS 1.2 [RFC6347], along with earlier protocol versions. 133 This guidance is included in the original [RFC7525] and mostly 134 retained in this revised version; note that this guidance was 135 mostly adopted by the industry since the publication of RFC 7525 136 in 2015. 138 * Versions of TLS earlier than 1.2 were deprecated [RFC8996]. 140 * Version 1.3 of TLS [RFC8446] was released and version 1.3 of DTLS 141 was finalized [I-D.ietf-tls-dtls13]; these versions largely 142 mitigate or resolve the described attacks. 144 Those who implement and deploy TLS and DTLS, in particular versions 145 1.2 or earlier of these protocols, need guidance on how TLS can be 146 used securely. This document provides guidance for deployed services 147 as well as for software implementations, assuming the implementer 148 expects his or her code to be deployed in environments defined in 149 Section 5. Concerning deployment, this document targets a wide 150 audience -- namely, all deployers who wish to add authentication (be 151 it one-way only or mutual), confidentiality, and data integrity 152 protection to their communications. 154 The recommendations herein take into consideration the security of 155 various mechanisms, their technical maturity and interoperability, 156 and their prevalence in implementations at the time of writing. 157 Unless it is explicitly called out that a recommendation applies to 158 TLS alone or to DTLS alone, each recommendation applies to both TLS 159 and DTLS. 161 This document attempts to minimize new guidance to TLS 1.2 162 implementations, and the overall approach is to encourage systems to 163 move to TLS 1.3. However this is not always practical. Newly 164 discovered attacks, as well as ecosystem changes, necessitated some 165 new requirements that apply to TLS 1.2 environments. Those are 166 summarized in Appendix A. 168 As noted, the TLS 1.3 specification resolves many of the 169 vulnerabilities listed in this document. A system that deploys TLS 170 1.3 should have fewer vulnerabilities than TLS 1.2 or below. This 171 document is being republished with this in mind, and with an explicit 172 goal to migrate most uses of TLS 1.2 into TLS 1.3. 174 These are minimum recommendations for the use of TLS in the vast 175 majority of implementation and deployment scenarios, with the 176 exception of unauthenticated TLS (see Section 5). Other 177 specifications that reference this document can have stricter 178 requirements related to one or more aspects of the protocol, based on 179 their particular circumstances (e.g., for use with a particular 180 application protocol); when that is the case, implementers are 181 advised to adhere to those stricter requirements. Furthermore, this 182 document provides a floor, not a ceiling, so stronger options are 183 always allowed (e.g., depending on differing evaluations of the 184 importance of cryptographic strength vs. computational load). 186 Community knowledge about the strength of various algorithms and 187 feasible attacks can change quickly, and experience shows that a Best 188 Current Practice (BCP) document about security is a point-in-time 189 statement. Readers are advised to seek out any errata or updates 190 that apply to this document. 192 2. Terminology 194 A number of security-related terms in this document are used in the 195 sense defined in [RFC4949]. 197 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 198 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 199 "OPTIONAL" in this document are to be interpreted as described in 200 BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all 201 capitals, as shown here. 203 3. General Recommendations 205 This section provides general recommendations on the secure use of 206 TLS. Recommendations related to cipher suites are discussed in the 207 following section. 209 3.1. Protocol Versions 211 3.1.1. SSL/TLS Protocol Versions 213 It is important both to stop using old, less secure versions of SSL/ 214 TLS and to start using modern, more secure versions; therefore, the 215 following are the recommendations concerning TLS/SSL protocol 216 versions: 218 * Implementations MUST NOT negotiate SSL version 2. 220 Rationale: Today, SSLv2 is considered insecure [RFC6176]. 222 * Implementations MUST NOT negotiate SSL version 3. 224 Rationale: SSLv3 [RFC6101] was an improvement over SSLv2 and 225 plugged some significant security holes but did not support strong 226 cipher suites. SSLv3 does not support TLS extensions, some of 227 which (e.g., renegotiation_info [RFC5746]) are security-critical. 228 In addition, with the emergence of the POODLE attack [POODLE], 229 SSLv3 is now widely recognized as fundamentally insecure. See 230 [DEP-SSLv3] for further details. 232 * Implementations MUST NOT negotiate TLS version 1.0 [RFC2246]. 234 Rationale: TLS 1.0 (published in 1999) does not support many 235 modern, strong cipher suites. In addition, TLS 1.0 lacks a per- 236 record Initialization Vector (IV) for CBC-based cipher suites and 237 does not warn against common padding errors. This and other 238 recommendations in this section are in line with [RFC8996]. 240 * Implementations MUST NOT negotiate TLS version 1.1 [RFC4346]. 242 Rationale: TLS 1.1 (published in 2006) is a security improvement 243 over TLS 1.0 but still does not support certain stronger cipher 244 suites. 246 * Implementations MUST support TLS 1.2 [RFC5246] and MUST prefer to 247 negotiate TLS version 1.2 over earlier versions of TLS. 249 Rationale: Several stronger cipher suites are available only with 250 TLS 1.2 (published in 2008). In fact, the cipher suites 251 recommended by this document for TLS 1.2 (Section 4.2 below) are 252 only available in this version. 254 * Implementations SHOULD support TLS 1.3 [RFC8446] and, if 255 implemented, MUST prefer to negotiate TLS 1.3 over earlier 256 versions of TLS. 258 Rationale: TLS 1.3 is a major overhaul to the protocol and 259 resolves many of the security issues with TLS 1.2. We note that 260 as long as TLS 1.2 is still allowed by a particular 261 implementation, even if it defaults to TLS 1.3, implementers MUST 262 still follow all the recommendations in this document. 264 * Implementations of "greenfield" protocols or deployments, where 265 there is no need to support legacy endpoints, SHOULD support TLS 266 1.3, with no negotiation of earlier versions. Similarly, we 267 RECOMMEND that new protocol designs that embed the TLS mechanisms 268 (such as QUIC has done [RFC9001]) include TLS 1.3. 270 Rationale: secure deployment of TLS 1.3 is significantly easier 271 and less error prone than the secure deployment of TLS 1.2. 273 This BCP applies to TLS 1.2, 1.3 and to earlier versions. It is not 274 safe for readers to assume that the recommendations in this BCP apply 275 to any future version of TLS. 277 3.1.2. DTLS Protocol Versions 279 DTLS, an adaptation of TLS for UDP datagrams, was introduced when TLS 280 1.1 was published. The following are the recommendations with 281 respect to DTLS: 283 * Implementations MUST NOT negotiate DTLS version 1.0 [RFC4347]. 285 Version 1.0 of DTLS correlates to version 1.1 of TLS (see above). 287 * Implementations MUST support DTLS 1.2 [RFC6347] and MUST prefer to 288 negotiate DTLS version 1.2 over earlier versions of DTLS. 290 Version 1.2 of DTLS correlates to version 1.2 of TLS (see above). 291 (There is no version 1.1 of DTLS.) 293 * Implementations SHOULD support DTLS 1.3 [I-D.ietf-tls-dtls13] and, 294 if implemented, MUST prefer to negotiate DTLS version 1.3 over 295 earlier versions of DTLS. 297 Version 1.3 of DTLS correlates to version 1.3 of TLS (see above). 299 3.1.3. Fallback to Lower Versions 301 TLS/DTLS 1.2 clients MUST NOT fall back to earlier TLS versions, 302 since those versions have been deprecated [RFC8996]. We note that as 303 a result of that, the SCSV mechanism [RFC7507] is no longer needed 304 for clients. In addition, TLS 1.3 implements a new version 305 negotiation mechanism. 307 3.2. Strict TLS 309 The following recommendations are provided to help prevent SSL 310 Stripping (an attack that is summarized in Section 2.1 of [RFC7457]): 312 * In cases where an application protocol allows implementations or 313 deployments a choice between strict TLS configuration and dynamic 314 upgrade from unencrypted to TLS-protected traffic (such as 315 STARTTLS), clients and servers SHOULD prefer strict TLS 316 configuration. 318 * Application protocols typically provide a way for the server to 319 offer TLS during an initial protocol exchange, and sometimes also 320 provide a way for the server to advertise support for TLS (e.g., 321 through a flag indicating that TLS is required); unfortunately, 322 these indications are sent before the communication channel is 323 encrypted. A client SHOULD attempt to negotiate TLS even if these 324 indications are not communicated by the server. 326 * HTTP client and server implementations MUST support the HTTP 327 Strict Transport Security (HSTS) header [RFC6797], in order to 328 allow Web servers to advertise that they are willing to accept 329 TLS-only clients. 331 * Web servers SHOULD use HSTS to indicate that they are willing to 332 accept TLS-only clients, unless they are deployed in such a way 333 that using HSTS would in fact weaken overall security (e.g., it 334 can be problematic to use HSTS with self-signed certificates, as 335 described in Section 11.3 of [RFC6797]). 337 Rationale: Combining unprotected and TLS-protected communication 338 opens the way to SSL Stripping and similar attacks, since an initial 339 part of the communication is not integrity protected and therefore 340 can be manipulated by an attacker whose goal is to keep the 341 communication in the clear. 343 3.3. Compression 345 In order to help prevent compression-related attacks (summarized in 346 Section 2.6 of [RFC7457]), when using TLS 1.2 implementations and 347 deployments SHOULD disable TLS-level compression (Section 6.2.2 of 348 [RFC5246]), unless the application protocol in question has been 349 shown not to be open to such attacks. Note: this recommendation 350 applies to TLS 1.2 only, because compression has been removed from 351 TLS 1.3. 353 Rationale: TLS compression has been subject to security attacks, such 354 as the CRIME attack. 356 Implementers should note that compression at higher protocol levels 357 can allow an active attacker to extract cleartext information from 358 the connection. The BREACH attack is one such case. These issues 359 can only be mitigated outside of TLS and are thus outside the scope 360 of this document. See Section 2.6 of [RFC7457] for further details. 362 3.4. TLS Session Resumption 364 Session resumption drastically reduces the number of TLS handshakes 365 and thus is an essential performance feature for most deployments. 367 Stateless session resumption with session tickets is a popular 368 strategy. For TLS 1.2, it is specified in [RFC5077]. For TLS 1.3, a 369 more secure PSK-based mechanism is described in Section 4.6.1 of 370 [RFC8446]. See this post (https://blog.filippo.io/we-need-to-talk- 371 about-session-tickets/) by Filippo Valsorda for a comparison of TLS 372 1.2 and 1.3 session resumption, and [Springall16] for a quantitative 373 study of TLS cryptographic "shortcuts", including session resumption. 375 When it is used, the resumption information MUST be authenticated and 376 encrypted to prevent modification or eavesdropping by an attacker. 377 Further recommendations apply to session tickets: 379 * A strong cipher suite MUST be used when encrypting the ticket (as 380 least as strong as the main TLS cipher suite). 382 * Ticket keys MUST be changed regularly, e.g., once every week, so 383 as not to negate the benefits of forward secrecy (see Section 6.3 384 for details on forward secrecy). Old ticket keys MUST be 385 destroyed shortly after a new key version is made available. 387 * For similar reasons, session ticket validity SHOULD be limited to 388 a reasonable duration (e.g., half as long as ticket key validity). 390 * TLS 1.2 does not roll the session key forward within a single 391 session. Thus, to prevent an attack where a stolen ticket key is 392 used to decrypt the entire content of a session (negating the 393 concept of forward secrecy), a TLS 1.2 server SHOULD NOT resume 394 sessions that are too old, e.g. sessions that have been open 395 longer than two ticket key rotation periods. Note that this 396 implies that some server implementations might need to abort 397 sessions after a certain duration. 399 Rationale: session resumption is another kind of TLS handshake, and 400 therefore must be as secure as the initial handshake. This document 401 (Section 4) recommends the use of cipher suites that provide forward 402 secrecy, i.e. that prevent an attacker who gains momentary access to 403 the TLS endpoint (either client or server) and its secrets from 404 reading either past or future communication. The tickets must be 405 managed so as not to negate this security property. 407 TLS 1.3 provides the powerful option of forward secrecy even within a 408 long-lived connection that is periodically resumed. Section 2.2 of 409 [RFC8446] recommends that clients SHOULD send a "key_share" when 410 initiating session resumption. In order to gain forward secrecy, 411 this document recommends that server implementations SHOULD respond 412 with a "key_share", to complete an ECDHE exchange on each session 413 resumption. 415 TLS session resumption introduces potential privacy issues where the 416 server is able to track the client, in some cases indefinitely. See 417 [Sy2018] for more details. 419 3.5. Renegotiation in TLS 1.2 421 The recommendations in this section apply to TLS 1.2 only, because 422 renegotiation has been removed from TLS 1.3. 424 TLS 1.2 clients and servers MUST implement the renegotiation_info 425 extension, as defined in [RFC5746]. 427 TLS 1.2 clients MUST send renegotiation_info in the Client Hello. If 428 the server does not acknowledge the extension, the client MUST 429 generate a fatal handshake_failure alert prior to terminating the 430 connection. 432 Rationale: It is not safe for a client to connect to a TLS 1.2 server 433 that does not support renegotiation_info, regardless of whether 434 either endpoint actually implements renegotiation. See also 435 Section 4.1 of [RFC5746]. 437 A related attack resulting from TLS session parameters not properly 438 authenticated is Triple Handshake [triple-handshake]. To address 439 this attack, TLS 1.2 implementations SHOULD support the 440 extended_master_secret extension defined in [RFC7627]. 442 3.6. Post-Handshake Authentication 444 Renegotiation in TLS 1.2 was replaced in TLS 1.3 by separate post- 445 handshake authentication and key update mechanisms. In the context 446 of protocols that multiplex requests over a single connection (such 447 as HTTP/2), post-handshake authentication has the same problems as 448 TLS 1.2 renegotiation. Multiplexed protocols SHOULD follow the 449 advice provided for HTTP/2 in [RFC8740]. 451 3.7. Server Name Indication 453 TLS implementations MUST support the Server Name Indication (SNI) 454 extension defined in Section 3 of [RFC6066] for those higher-level 455 protocols that would benefit from it, including HTTPS. However, the 456 actual use of SNI in particular circumstances is a matter of local 457 policy. Implementers are strongly encouraged to support TLS 458 Encrypted Client Hello (formerly called Encrypted SNI) once 459 [I-D.ietf-tls-esni] has been standardized. 461 Rationale: SNI supports deployment of multiple TLS-protected virtual 462 servers on a single address, and therefore enables fine-grained 463 security for these virtual servers, by allowing each one to have its 464 own certificate. However, SNI also leaks the target domain for a 465 given connection; this information leak will be plugged by use of TLS 466 Encrypted Client Hello. 468 In order to prevent the attacks described in [ALPACA], a server that 469 does not recognize the presented server name SHOULD NOT continue the 470 handshake and instead SHOULD fail with a fatal-level 471 unrecognized_name(112) alert. Note that this recommendation updates 472 Section 3 of [RFC6066]: "If the server understood the ClientHello 473 extension but does not recognize the server name, the server SHOULD 474 take one of two actions: either abort the handshake by sending a 475 fatal-level unrecognized_name(112) alert or continue the handshake." 476 It is also RECOMMENDED that clients abort the handshake if the server 477 acknowledges the SNI extension, but presents a certificate with a 478 different hostname than the one sent by the client. 480 3.8. Application-Layer Protocol Negotiation 482 TLS implementations (both client- and server-side) MUST support the 483 Application-Layer Protocol Negotiation (ALPN) extension [RFC7301]. 485 In order to prevent "cross-protocol" attacks resulting from failure 486 to ensure that a message intended for use in one protocol cannot be 487 mistaken for a message for use in another protocol, servers should 488 strictly enforce the behavior prescribed in Section 3.2 of [RFC7301]: 489 "In the event that the server supports no protocols that the client 490 advertises, then the server SHALL respond with a fatal 491 no_application_protocol alert." It is also RECOMMENDED that clients 492 abort the handshake if the server acknowledges the ALPN extension, 493 but does not select a protocol from the client list. Failure to do 494 so can result in attacks such those described in [ALPACA]. 496 Protocol developers are strongly encouraged to register an ALPN 497 identifier for their protocols. This applies to new protocols, as 498 well as well-established protocols such as SMTP. 500 3.9. Zero Round Trip Time (0-RTT) Data in TLS 1.3 502 The 0-RTT early data feature is new in TLS 1.3. It provides improved 503 latency when TLS connections are resumed, at the potential cost of 504 security. As a result, it requires special attention from 505 implementers on both the server and the client side. Typically this 506 extends to both the TLS library as well as protocol layers above it. 508 For use in HTTP-over-TLS, readers are referred to [RFC8470] for 509 guidance. 511 For QUIC-on-TLS, refer to Sec. 9.2 of [RFC9001]. 513 For other protocols, generic guidance is given in Sec. 8 and 514 Appendix E.5 of [RFC8446]. To paraphrase Appendix E.5, applications 515 MUST avoid this feature unless an explicit specification exists for 516 the application protocol in question to clarify when 0-RTT is 517 appropriate and secure. This can take the form of an IETF RFC, a 518 non-IETF standard, or even documentation associated with a non- 519 standard protocol. 521 4. Recommendations: Cipher Suites 523 TLS and its implementations provide considerable flexibility in the 524 selection of cipher suites. Unfortunately, the security of some of 525 these cipher suites has degraded over time to the point where some 526 are known to be insecure. Incorrectly configuring a server leads to 527 no or reduced security. This section includes recommendations on the 528 selection and negotiation of cipher suites. 530 4.1. General Guidelines 532 Cryptographic algorithms weaken over time as cryptanalysis improves: 533 algorithms that were once considered strong become weak. Such 534 algorithms need to be phased out over time and replaced with more 535 secure cipher suites. This helps to ensure that the desired security 536 properties still hold. SSL/TLS has been in existence for almost 20 537 years and many of the cipher suites that have been recommended in 538 various versions of SSL/TLS are now considered weak or at least not 539 as strong as desired. Therefore, this section modernizes the 540 recommendations concerning cipher suite selection. 542 * Implementations MUST NOT negotiate the cipher suites with NULL 543 encryption. 545 Rationale: The NULL cipher suites do not encrypt traffic and so 546 provide no confidentiality services. Any entity in the network 547 with access to the connection can view the plaintext of contents 548 being exchanged by the client and server. 549 Nevertheless, this document does not discourage software from 550 implementing NULL cipher suites, since they can be useful for 551 testing and debugging. 553 * Implementations MUST NOT negotiate RC4 cipher suites. 555 Rationale: The RC4 stream cipher has a variety of cryptographic 556 weaknesses, as documented in [RFC7465]. Note that DTLS 557 specifically forbids the use of RC4 already. 559 * Implementations MUST NOT negotiate cipher suites offering less 560 than 112 bits of security, including so-called "export-level" 561 encryption (which provide 40 or 56 bits of security). 563 Rationale: Based on [RFC3766], at least 112 bits of security is 564 needed. 40-bit and 56-bit security are considered insecure today. 565 TLS 1.1 and 1.2 never negotiate 40-bit or 56-bit export ciphers. 567 * Implementations SHOULD NOT negotiate cipher suites that use 568 algorithms offering less than 128 bits of security. 570 Rationale: Cipher suites that offer between 112-bits and 128-bits 571 of security are not considered weak at this time; however, it is 572 expected that their useful lifespan is short enough to justify 573 supporting stronger cipher suites at this time. 128-bit ciphers 574 are expected to remain secure for at least several years, and 575 256-bit ciphers until the next fundamental technology 576 breakthrough. Note that, because of so-called "meet-in-the- 577 middle" attacks [Multiple-Encryption], some legacy cipher suites 578 (e.g., 168-bit 3DES) have an effective key length that is smaller 579 than their nominal key length (112 bits in the case of 3DES). 580 Such cipher suites should be evaluated according to their 581 effective key length. 583 * Implementations SHOULD NOT negotiate cipher suites based on RSA 584 key transport, a.k.a. "static RSA". 586 Rationale: These cipher suites, which have assigned values 587 starting with the string "TLS_RSA_WITH_*", have several drawbacks, 588 especially the fact that they do not support forward secrecy. 590 * Implementations SHOULD NOT negotiate cipher suites based on non- 591 ephemeral (static) finite-field Diffie-Hellman key agreement. 593 Rationale: These cipher suites, which have assigned values 594 prefixed by "TLS_DH_*", have several drawbacks, especially the 595 fact that they do not support forward secrecy. 597 * Implementations MUST support and prefer to negotiate cipher suites 598 offering forward secrecy. However, TLS 1.2 implementations SHOULD 599 NOT negotiate cipher suites based on ephemeral finite-field 600 Diffie-Hellman key agreement (i.e., "TLS_DHE_*" suites). This is 601 justified by the known fragility of the construction (see 602 [RACCOON]) and the limitation around negotiation, including using 603 [RFC7919], which has seen very limited uptake. 605 Rationale: Forward secrecy (sometimes called "perfect forward 606 secrecy") prevents the recovery of information that was encrypted 607 with older session keys, thus limiting the amount of time during 608 which attacks can be successful. See Section 6.3 for a detailed 609 discussion. 611 4.2. Cipher Suites for TLS 1.2 613 Given the foregoing considerations, implementation and deployment of 614 the following cipher suites is RECOMMENDED: 616 * TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 617 * TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384 619 * TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256 621 * TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384 623 These cipher suites are supported only in TLS 1.2 and not in earlier 624 protocol versions, because they are authenticated encryption (AEAD) 625 algorithms [RFC5116]. 627 Typically, in order to prefer these suites, the order of suites needs 628 to be explicitly configured in server software. (See [BETTERCRYPTO] 629 for helpful deployment guidelines, but note that its recommendations 630 differ from the current document in some details.) It would be ideal 631 if server software implementations were to prefer these suites by 632 default. 634 Some devices have hardware support for AES-CCM but not AES-GCM, so 635 they are unable to follow the foregoing recommendations regarding 636 cipher suites. There are even devices that do not support public key 637 cryptography at all, but they are out of scope entirely. 639 When using ECDSA signatures for authentication of TLS peers, it is 640 RECOMMENDED that implementations use the NIST curve P-256. In 641 addition, to avoid predictable or repeated nonces (that would allow 642 revealing the long term signing key), it is RECOMMENDED that 643 implementations implement "deterministic ECDSA" as specified in 644 [RFC6979] and in line with the recommendations in [RFC8446]. 646 4.2.1. Implementation Details 648 Clients SHOULD include TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256 as the 649 first proposal to any server, unless they have prior knowledge that 650 the server cannot respond to a TLS 1.2 client_hello message. 652 Servers MUST prefer this cipher suite over weaker cipher suites 653 whenever it is proposed, even if it is not the first proposal. 655 Clients are of course free to offer stronger cipher suites, e.g., 656 using AES-256; when they do, the server SHOULD prefer the stronger 657 cipher suite unless there are compelling reasons (e.g., seriously 658 degraded performance) to choose otherwise. 660 The previous version of this document implicitly allowed the old RFC 661 5246 mandatory-to-implement cipher suite, 662 TLS_RSA_WITH_AES_128_CBC_SHA. At the time of writing, this cipher 663 suite does not provide additional interoperability, except with 664 extremely old clients. As with other cipher suites that do not 665 provide forward secrecy, implementations SHOULD NOT support this 666 cipher suite. Other application protocols specify other cipher 667 suites as mandatory to implement (MTI). 669 [RFC8422] allows clients and servers to negotiate ECDH parameters 670 (curves). Both clients and servers SHOULD include the "Supported 671 Elliptic Curves" extension [RFC8422]. Clients and servers SHOULD 672 support the NIST P-256 (secp256r1) [RFC8422] and X25519 (x25519) 673 [RFC7748] curves. Note that [RFC8422] deprecates all but the 674 uncompressed point format. Therefore, if the client sends an 675 ec_point_formats extension, the ECPointFormatList MUST contain a 676 single element, "uncompressed". 678 4.3. Cipher Suites for TLS 1.3 680 This document does not specify any cipher suites for TLS 1.3. 681 Readers are referred to Sec. 9.1 of [RFC8446] for cipher suite 682 recommendations. 684 4.4. Limits on Key Usage 686 All ciphers have an upper limit on the amount of traffic that can be 687 securely protected with any given key. In the case of AEAD cipher 688 suites, two separate limits are maintained for each key: 690 1. Confidentiality limit (CL), i.e., the number of records that can 691 be encrypted. 693 2. Integrity limit (IL), i.e., the number of records that are 694 allowed to fail authentication. 696 The latter only applies to DTLS since TLS connections are torn down 697 on the first decryption failure. 699 When a sender is approaching CL, the implementation SHOULD initiate a 700 new handshake (or in TLS 1.3, a Key Update) to rotate the session 701 key. 703 When a receiver has reached IL, the implementation SHOULD close the 704 connection. 706 For all TLS 1.3 cipher suites, readers are referred to Section 5.5 of 707 [RFC8446] for the values of CL and IL. For all DTLS 1.3 cipher 708 suites, readers are referred to Section 4.5.3 of 709 [I-D.ietf-tls-dtls13]. 711 For all AES-GCM cipher suites recommended for TLS 1.2 and DTLS 1.2 in 712 this document, CL can be derived by plugging the corresponding 713 parameters into the inequalities in Section 6.1 of 714 [I-D.irtf-cfrg-aead-limits] that apply to random, partially implicit 715 nonces, i.e., the nonce construction used in TLS 1.2. Although the 716 obtained figures are slightly higher than those for TLS 1.3, it is 717 RECOMMENDED that the same limit of 2^24.5 records is used for both 718 versions. 720 For all AES-GCM cipher suites recommended for DTLS 1.2, IL (obtained 721 from the same inequalities referenced above) is 2^28. 723 4.5. Public Key Length 725 When using the cipher suites recommended in this document, two public 726 keys are normally used in the TLS handshake: one for the Diffie- 727 Hellman key agreement and one for server authentication. Where a 728 client certificate is used, a third public key is added. 730 With a key exchange based on modular exponential (MODP) Diffie- 731 Hellman groups ("DHE" cipher suites), DH key lengths of at least 2048 732 bits are REQUIRED. 734 Rationale: For various reasons, in practice, DH keys are typically 735 generated in lengths that are powers of two (e.g., 2^10 = 1024 bits, 736 2^11 = 2048 bits, 2^12 = 4096 bits). Because a DH key of 1228 bits 737 would be roughly equivalent to only an 80-bit symmetric key 738 [RFC3766], it is better to use keys longer than that for the "DHE" 739 family of cipher suites. A DH key of 1926 bits would be roughly 740 equivalent to a 100-bit symmetric key [RFC3766]. A DH key of 2048 741 bits (equivalent to a 112-bit symmetric key) is the minimum allowed 742 by the latest revision of [NIST.SP.800-56A], as of this writing (see 743 in particular Appendix D). 745 As noted in [RFC3766], correcting for the emergence of a TWIRL 746 machine would imply that 1024-bit DH keys yield about 65 bits of 747 equivalent strength and that a 2048-bit DH key would yield about 92 748 bits of equivalent strength. The Logjam attack [Logjam] further 749 demonstrates that 1024-bit Diffie Hellman parameters should be 750 avoided. 752 With regard to ECDH keys, implementers are referred to the IANA 753 "Supported Groups Registry" (former "EC Named Curve Registry"), 754 within the "Transport Layer Security (TLS) Parameters" registry 755 [IANA_TLS], and in particular to the "recommended" groups. Curves of 756 less than 224 bits MUST NOT be used. This recommendation is in-line 757 with the latest revision of [NIST.SP.800-56A]. 759 When using RSA, servers SHOULD authenticate using certificates with 760 at least a 2048-bit modulus for the public key. In addition, the use 761 of the SHA-256 hash algorithm is RECOMMENDED and SHA-1 or MD5 MUST 762 NOT be used ([RFC9155], and see [CAB-Baseline] for more details). 763 Clients MUST indicate to servers that they request SHA-256, by using 764 the "Signature Algorithms" extension defined in TLS 1.2. 766 4.6. Truncated HMAC 768 Implementations MUST NOT use the Truncated HMAC extension, defined in 769 Section 7 of [RFC6066]. 771 Rationale: the extension does not apply to the AEAD cipher suites 772 recommended above. However it does apply to most other TLS cipher 773 suites. Its use has been shown to be insecure in [PatersonRS11]. 775 5. Applicability Statement 777 The recommendations of this document primarily apply to the 778 implementation and deployment of application protocols that are most 779 commonly used with TLS and DTLS on the Internet today. Examples 780 include, but are not limited to: 782 * Web software and services that wish to protect HTTP traffic with 783 TLS. 785 * Email software and services that wish to protect IMAP, POP3, or 786 SMTP traffic with TLS. 788 * Instant-messaging software and services that wish to protect 789 Extensible Messaging and Presence Protocol (XMPP) or Internet 790 Relay Chat (IRC) traffic with TLS. 792 * Realtime media software and services that wish to protect Secure 793 Realtime Transport Protocol (SRTP) traffic with DTLS. 795 This document does not modify the implementation and deployment 796 recommendations (e.g., mandatory-to-implement cipher suites) 797 prescribed by existing application protocols that employ TLS or DTLS. 798 If the community that uses such an application protocol wishes to 799 modernize its usage of TLS or DTLS to be consistent with the best 800 practices recommended here, it needs to explicitly update the 801 existing application protocol definition (one example is [RFC7590], 802 which updates [RFC6120]). 804 Designers of new application protocols developed through the Internet 805 Standards Process [RFC2026] are expected at minimum to conform to the 806 best practices recommended here, unless they provide documentation of 807 compelling reasons that would prevent such conformance (e.g., 808 widespread deployment on constrained devices that lack support for 809 the necessary algorithms). 811 5.1. Security Services 813 This document provides recommendations for an audience that wishes to 814 secure their communication with TLS to achieve the following: 816 * Confidentiality: all application-layer communication is encrypted 817 with the goal that no party should be able to decrypt it except 818 the intended receiver. 820 * Data integrity: any changes made to the communication in transit 821 are detectable by the receiver. 823 * Authentication: an endpoint of the TLS communication is 824 authenticated as the intended entity to communicate with. 826 With regard to authentication, TLS enables authentication of one or 827 both endpoints in the communication. In the context of opportunistic 828 security [RFC7435], TLS is sometimes used without authentication. As 829 discussed in Section 5.2, considerations for opportunistic security 830 are not in scope for this document. 832 If deployers deviate from the recommendations given in this document, 833 they need to be aware that they might lose access to one of the 834 foregoing security services. 836 This document applies only to environments where confidentiality is 837 required. It recommends algorithms and configuration options that 838 enforce secrecy of the data in transit. 840 This document also assumes that data integrity protection is always 841 one of the goals of a deployment. In cases where integrity is not 842 required, it does not make sense to employ TLS in the first place. 843 There are attacks against confidentiality-only protection that 844 utilize the lack of integrity to also break confidentiality (see, for 845 instance, [DegabrieleP07] in the context of IPsec). 847 This document addresses itself to application protocols that are most 848 commonly used on the Internet with TLS and DTLS. Typically, all 849 communication between TLS clients and TLS servers requires all three 850 of the above security services. This is particularly true where TLS 851 clients are user agents like Web browsers or email software. 853 This document does not address the rarer deployment scenarios where 854 one of the above three properties is not desired, such as the use 855 case described in Section 5.2 below. As another scenario where 856 confidentiality is not needed, consider a monitored network where the 857 authorities in charge of the respective traffic domain require full 858 access to unencrypted (plaintext) traffic, and where users 859 collaborate and send their traffic in the clear. 861 5.2. Opportunistic Security 863 There are several important scenarios in which the use of TLS is 864 optional, i.e., the client decides dynamically ("opportunistically") 865 whether to use TLS with a particular server or to connect in the 866 clear. This practice, often called "opportunistic security", is 867 described at length in [RFC7435] and is often motivated by a desire 868 for backward compatibility with legacy deployments. 870 In these scenarios, some of the recommendations in this document 871 might be too strict, since adhering to them could cause fallback to 872 cleartext, a worse outcome than using TLS with an outdated protocol 873 version or cipher suite. 875 6. Security Considerations 877 This entire document discusses the security practices directly 878 affecting applications using the TLS protocol. This section contains 879 broader security considerations related to technologies used in 880 conjunction with or by TLS. 882 6.1. Host Name Validation 884 Application authors should take note that some TLS implementations do 885 not validate host names. If the TLS implementation they are using 886 does not validate host names, authors might need to write their own 887 validation code or consider using a different TLS implementation. 889 It is noted that the requirements regarding host name validation 890 (and, in general, binding between the TLS layer and the protocol that 891 runs above it) vary between different protocols. For HTTPS, these 892 requirements are defined by Sections 4.3.3, 4.3.4 and 4.3.5 of 893 [I-D.ietf-httpbis-semantics]. 895 Readers are referred to [RFC6125] for further details regarding 896 generic host name validation in the TLS context. In addition, that 897 RFC contains a long list of example protocols, some of which 898 implement a policy very different from HTTPS. 900 If the host name is discovered indirectly and in an insecure manner 901 (e.g., by an insecure DNS query for an MX or SRV record), it SHOULD 902 NOT be used as a reference identifier [RFC6125] even when it matches 903 the presented certificate. This proviso does not apply if the host 904 name is discovered securely (for further discussion, see [DANE-SRV] 905 and [DANE-SMTP]). 907 Host name validation typically applies only to the leaf "end entity" 908 certificate. Naturally, in order to ensure proper authentication in 909 the context of the PKI, application clients need to verify the entire 910 certification path in accordance with [RFC5280] (see also [RFC6125]). 912 6.2. AES-GCM 914 Section 4.2 above recommends the use of the AES-GCM authenticated 915 encryption algorithm. Please refer to Section 11 of [RFC5246] for 916 general security considerations when using TLS 1.2, and to Section 6 917 of [RFC5288] for security considerations that apply specifically to 918 AES-GCM when used with TLS. 920 6.2.1. Nonce Reuse in TLS 1.2 922 The existence of deployed TLS stacks that mistakenly reuse the AES- 923 GCM nonce is documented in [Boeck2016], showing there is an actual 924 risk of AES-GCM getting implemented in an insecure way and thus 925 making TLS sessions that use an AES-GCM cipher suite vulnerable to 926 attacks such as [Joux2006]. (See [CVE] records: CVE-2016-0270, CVE- 927 2016-10213, CVE-2016-10212, CVE-2017-5933.) 929 While this problem has been fixed in TLS 1.3, which enforces a 930 deterministic method to generate nonces from record sequence numbers 931 and shared secrets for all of its AEAD cipher suites (including AES- 932 GCM), TLS 1.2 implementations could still choose their own 933 (potentially insecure) nonce generation methods. 935 It is therefore RECOMMENDED that TLS 1.2 implementations use the 936 64-bit sequence number to populate the nonce_explicit part of the GCM 937 nonce, as described in the first two paragraphs of Section 5.3 of 938 [RFC8446]. Note that this recommendation updates Section 3 of 939 [RFC5288]: "The nonce_explicit MAY be the 64-bit sequence number." 941 We note that at the time of writing there are no cipher suites 942 defined for nonce reuse resistant algorithms such as AES-GCM-SIV 943 [RFC8452]. 945 6.3. Forward Secrecy 947 Forward secrecy (also called "perfect forward secrecy" or "PFS" and 948 defined in [RFC4949]) is a defense against an attacker who records 949 encrypted conversations where the session keys are only encrypted 950 with the communicating parties' long-term keys. 952 Should the attacker be able to obtain these long-term keys at some 953 point later in time, the session keys and thus the entire 954 conversation could be decrypted. 956 In the context of TLS and DTLS, such compromise of long-term keys is 957 not entirely implausible. It can happen, for example, due to: 959 * A client or server being attacked by some other attack vector, and 960 the private key retrieved. 962 * A long-term key retrieved from a device that has been sold or 963 otherwise decommissioned without prior wiping. 965 * A long-term key used on a device as a default key [Heninger2012]. 967 * A key generated by a trusted third party like a CA, and later 968 retrieved from it either by extortion or compromise 969 [Soghoian2011]. 971 * A cryptographic break-through, or the use of asymmetric keys with 972 insufficient length [Kleinjung2010]. 974 * Social engineering attacks against system administrators. 976 * Collection of private keys from inadequately protected backups. 978 Forward secrecy ensures in such cases that it is not feasible for an 979 attacker to determine the session keys even if the attacker has 980 obtained the long-term keys some time after the conversation. It 981 also protects against an attacker who is in possession of the long- 982 term keys but remains passive during the conversation. 984 Forward secrecy is generally achieved by using the Diffie-Hellman 985 scheme to derive session keys. The Diffie-Hellman scheme has both 986 parties maintain private secrets and send parameters over the network 987 as modular powers over certain cyclic groups. The properties of the 988 so-called Discrete Logarithm Problem (DLP) allow the parties to 989 derive the session keys without an eavesdropper being able to do so. 990 There is currently no known attack against DLP if sufficiently large 991 parameters are chosen. A variant of the Diffie-Hellman scheme uses 992 elliptic curves instead of the originally proposed modular 993 arithmetic. Given the current state of the art, elliptic-curve 994 Diffie-Hellman appears to be more efficient, permits shorter key 995 lengths, and allows less freedom for implementation errors than 996 finite-field Diffie-Hellman. 998 Unfortunately, many TLS/DTLS cipher suites were defined that do not 999 feature forward secrecy, e.g., TLS_RSA_WITH_AES_256_CBC_SHA256. This 1000 document therefore advocates strict use of forward-secrecy-only 1001 ciphers. 1003 6.4. Diffie-Hellman Exponent Reuse 1005 For performance reasons, many TLS implementations reuse Diffie- 1006 Hellman and Elliptic Curve Diffie-Hellman exponents across multiple 1007 connections. Such reuse can result in major security issues: 1009 * If exponents are reused for too long (in some cases, even as 1010 little as a few hours), an attacker who gains access to the host 1011 can decrypt previous connections. In other words, exponent reuse 1012 negates the effects of forward secrecy. 1014 * TLS implementations that reuse exponents should test the DH public 1015 key they receive for group membership, in order to avoid some 1016 known attacks. These tests are not standardized in TLS at the 1017 time of writing, although general guidance in this area is 1018 provided by [NIST.SP.800-56A] and available in many protocol 1019 implementations. 1021 * Under certain conditions, the use of static finite-field DH keys, 1022 or of ephemeral finite-field DH keys that are reused across 1023 multiple connections, can lead to timing attacks (such as those 1024 described in [RACCOON]) on the shared secrets used in Diffie- 1025 Hellman key exchange. 1027 * An "invalid curve" attack can be mounted against elliptic-curve DH 1028 if the victim does not verify that the received point lies on the 1029 correct curve. If the victim is reusing the DH secrets, the 1030 attacker can repeat the probe varying the points to recover the 1031 full secret (see [Antipa2003] and [Jager2015]). 1033 To address these concerns: 1035 * TLS implementations SHOULD NOT use static finite-field DH keys and 1036 SHOULD NOT reuse ephemeral finite-field DH keys across multiple 1037 connections. 1039 * Server implementations that want to reuse elliptic-curve DH keys 1040 SHOULD either use a "safe curve" [SAFECURVES] (e.g., X25519), or 1041 perform the checks described in [NIST.SP.800-56A] on the received 1042 points. 1044 6.5. Certificate Revocation 1046 The following considerations and recommendations represent the 1047 current state of the art regarding certificate revocation, even 1048 though no complete and efficient solution exists for the problem of 1049 checking the revocation status of common public key certificates 1050 [RFC5280]: 1052 * Certificate revocation is an important tool when recovering from 1053 attacks on the TLS implementation, as well as cases of misissued 1054 certificates. TLS implementations MUST implement a strategy to 1055 distrust revoked certificates. 1057 * Although Certificate Revocation Lists (CRLs) are the most widely 1058 supported mechanism for distributing revocation information, they 1059 have known scaling challenges that limit their usefulness, despite 1060 workarounds such as partitioned CRLs and delta CRLs. The more 1061 modern [CRLite] and the follow-on Let's Revoke [LetsRevoke] build 1062 on the availability of Certificate Transparency [RFC9162] logs and 1063 aggressive compression to allow practical use of the CRL 1064 infrastructure, but at the time of writing, neither solution is 1065 deployed for client-side revocation processing at scale. 1067 * Proprietary mechanisms that embed revocation lists in the Web 1068 browser's configuration database cannot scale beyond a small 1069 number of the most heavily used Web servers. 1071 * The On-Line Certification Status Protocol (OCSP) [RFC6960] in its 1072 basic form presents both scaling and privacy issues. In addition, 1073 clients typically "soft-fail", meaning that they do not abort the 1074 TLS connection if the OCSP server does not respond. (However, 1075 this might be a workaround to avoid denial-of-service attacks if 1076 an OCSP responder is taken offline.). For an up-to-date survey of 1077 the status of OCSP deployment in the Web PKI see [Chung18]. 1079 * The TLS Certificate Status Request extension (Section 8 of 1080 [RFC6066]), commonly called "OCSP stapling", resolves the 1081 operational issues with OCSP. However, it is still ineffective in 1082 the presence of a MITM attacker because the attacker can simply 1083 ignore the client's request for a stapled OCSP response. 1085 * [RFC7633] defines a certificate extension that indicates that 1086 clients must expect stapled OCSP responses for the certificate and 1087 must abort the handshake ("hard-fail") if such a response is not 1088 available. 1090 * OCSP stapling as used in TLS 1.2 does not extend to intermediate 1091 certificates within a certificate chain. The Multiple Certificate 1092 Status extension [RFC6961] addresses this shortcoming, but it has 1093 seen little deployment and had been deprecated by [RFC8446]. As a 1094 result, we no longer recommend this extension for TLS 1.2. 1096 * TLS 1.3 (Section 4.4.2.1 of [RFC8446]) allows the association of 1097 OCSP information with intermediate certificates by using an 1098 extension to the CertificateEntry structure. However using this 1099 facility remains impractical because many CAs either do not 1100 publish OCSP for CA certificates or publish OCSP reports with a 1101 lifetime that is too long to be useful. 1103 * Both CRLs and OCSP depend on relatively reliable connectivity to 1104 the Internet, which might not be available to certain kinds of 1105 nodes. A common example is newly provisioned devices that need to 1106 establish a secure connection in order to boot up for the first 1107 time. 1109 For the common use cases of public key certificates in TLS, servers 1110 SHOULD support the following as a best practice given the current 1111 state of the art and as a foundation for a possible future solution: 1112 OCSP [RFC6960] and OCSP stapling using the status_request extension 1113 defined in [RFC6066]. Note that the exact mechanism for embedding 1114 the status_request extension differs between TLS 1.2 and 1.3. As a 1115 matter of local policy, server operators MAY request that CAs issue 1116 must-staple [RFC7633] certificates for the server and/or for client 1117 authentication, but we recommend to review the operational conditions 1118 before deciding on this approach. 1120 The considerations in this section do not apply to scenarios where 1121 the DANE-TLSA resource record [RFC6698] is used to signal to a client 1122 which certificate a server considers valid and good to use for TLS 1123 connections. 1125 7. Acknowledgments 1127 Thanks to Alexey Melnikov, Andrei Popov, Christian Huitema, Daniel 1128 Kahn Gillmor, David Benjamin, Eric Rescorla, Hannes Tschofenig, 1129 Hubert Kario, Ilari Liusvaara, John Mattsson, John R Levine, Julien 1130 Élie, Martin Thomson, Mohit Sahni, Nick Sullivan, Nimrod Aviram, Rich 1131 Salz, Ryan Sleevi, Sean Turner, Valery Smyslov, Viktor Dukhovni for 1132 helpful comments and discussions that have shaped this document. 1134 The authors gratefully acknowledge the contribution of Ralph Holz, 1135 who was a coauthor of RFC 7525, the previous version of this 1136 document. 1138 See RFC 7525 for additional acknowledgments for the previous revision 1139 of this document. 1141 8. References 1143 8.1. Normative References 1145 [I-D.ietf-httpbis-semantics] 1146 Fielding, R. T., Nottingham, M., and J. Reschke, "HTTP 1147 Semantics", Work in Progress, Internet-Draft, draft-ietf- 1148 httpbis-semantics-19, 12 September 2021, 1149 . 1152 [I-D.ietf-tls-dtls13] 1153 Rescorla, E., Tschofenig, H., and N. Modadugu, "The 1154 Datagram Transport Layer Security (DTLS) Protocol Version 1155 1.3", Work in Progress, Internet-Draft, draft-ietf-tls- 1156 dtls13-43, 30 April 2021, 1157 . 1160 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 1161 Requirement Levels", BCP 14, RFC 2119, 1162 DOI 10.17487/RFC2119, March 1997, 1163 . 1165 [RFC3766] Orman, H. and P. Hoffman, "Determining Strengths For 1166 Public Keys Used For Exchanging Symmetric Keys", BCP 86, 1167 RFC 3766, DOI 10.17487/RFC3766, April 2004, 1168 . 1170 [RFC4949] Shirey, R., "Internet Security Glossary, Version 2", 1171 FYI 36, RFC 4949, DOI 10.17487/RFC4949, August 2007, 1172 . 1174 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 1175 (TLS) Protocol Version 1.2", RFC 5246, 1176 DOI 10.17487/RFC5246, August 2008, 1177 . 1179 [RFC5288] Salowey, J., Choudhury, A., and D. McGrew, "AES Galois 1180 Counter Mode (GCM) Cipher Suites for TLS", RFC 5288, 1181 DOI 10.17487/RFC5288, August 2008, 1182 . 1184 [RFC5746] Rescorla, E., Ray, M., Dispensa, S., and N. Oskov, 1185 "Transport Layer Security (TLS) Renegotiation Indication 1186 Extension", RFC 5746, DOI 10.17487/RFC5746, February 2010, 1187 . 1189 [RFC6066] Eastlake 3rd, D., "Transport Layer Security (TLS) 1190 Extensions: Extension Definitions", RFC 6066, 1191 DOI 10.17487/RFC6066, January 2011, 1192 . 1194 [RFC6125] Saint-Andre, P. and J. Hodges, "Representation and 1195 Verification of Domain-Based Application Service Identity 1196 within Internet Public Key Infrastructure Using X.509 1197 (PKIX) Certificates in the Context of Transport Layer 1198 Security (TLS)", RFC 6125, DOI 10.17487/RFC6125, March 1199 2011, . 1201 [RFC6176] Turner, S. and T. Polk, "Prohibiting Secure Sockets Layer 1202 (SSL) Version 2.0", RFC 6176, DOI 10.17487/RFC6176, March 1203 2011, . 1205 [RFC6347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1206 Security Version 1.2", RFC 6347, DOI 10.17487/RFC6347, 1207 January 2012, . 1209 [RFC6979] Pornin, T., "Deterministic Usage of the Digital Signature 1210 Algorithm (DSA) and Elliptic Curve Digital Signature 1211 Algorithm (ECDSA)", RFC 6979, DOI 10.17487/RFC6979, August 1212 2013, . 1214 [RFC7301] Friedl, S., Popov, A., Langley, A., and E. Stephan, 1215 "Transport Layer Security (TLS) Application-Layer Protocol 1216 Negotiation Extension", RFC 7301, DOI 10.17487/RFC7301, 1217 July 2014, . 1219 [RFC7465] Popov, A., "Prohibiting RC4 Cipher Suites", RFC 7465, 1220 DOI 10.17487/RFC7465, February 2015, 1221 . 1223 [RFC7627] Bhargavan, K., Ed., Delignat-Lavaud, A., Pironti, A., 1224 Langley, A., and M. Ray, "Transport Layer Security (TLS) 1225 Session Hash and Extended Master Secret Extension", 1226 RFC 7627, DOI 10.17487/RFC7627, September 2015, 1227 . 1229 [RFC7748] Langley, A., Hamburg, M., and S. Turner, "Elliptic Curves 1230 for Security", RFC 7748, DOI 10.17487/RFC7748, January 1231 2016, . 1233 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 1234 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 1235 May 2017, . 1237 [RFC8422] Nir, Y., Josefsson, S., and M. Pegourie-Gonnard, "Elliptic 1238 Curve Cryptography (ECC) Cipher Suites for Transport Layer 1239 Security (TLS) Versions 1.2 and Earlier", RFC 8422, 1240 DOI 10.17487/RFC8422, August 2018, 1241 . 1243 [RFC8446] Rescorla, E., "The Transport Layer Security (TLS) Protocol 1244 Version 1.3", RFC 8446, DOI 10.17487/RFC8446, August 2018, 1245 . 1247 [RFC8740] Benjamin, D., "Using TLS 1.3 with HTTP/2", RFC 8740, 1248 DOI 10.17487/RFC8740, February 2020, 1249 . 1251 [RFC8996] Moriarty, K. and S. Farrell, "Deprecating TLS 1.0 and TLS 1252 1.1", BCP 195, RFC 8996, DOI 10.17487/RFC8996, March 2021, 1253 . 1255 [RFC9155] Velvindron, L., Moriarty, K., and A. Ghedini, "Deprecating 1256 MD5 and SHA-1 Signature Hashes in TLS 1.2 and DTLS 1.2", 1257 RFC 9155, DOI 10.17487/RFC9155, December 2021, 1258 . 1260 8.2. Informative References 1262 [ALPACA] Brinkmann, M., Dresen, C., Merget, R., Poddebniak, D., 1263 Müller, J., Somorovsky, J., Schwenk, J., and S. Schinzel, 1264 "ALPACA: Application Layer Protocol Confusion - Analyzing 1265 and Mitigating Cracks in TLS Authentication", 30th USENIX 1266 Security Symposium (USENIX Security 21) , 2021, 1267 . 1270 [Antipa2003] 1271 Antipa, A., Brown, D.R.L., Menezes, A., Struik, R., and 1272 S.A. Vanstone, "Validation of Elliptic Curve Public Keys", 1273 Public Key Cryptography - PKC 2003 , 2003. 1275 [BETTERCRYPTO] 1276 bettercrypto.org, "Applied Crypto Hardening", April 2015, 1277 . 1279 [Boeck2016] 1280 Böck, H., Zauner, A., Devlin, S., Somorovsky, J., and P. 1281 Jovanovic, "Nonce-Disrespecting Adversaries: Practical 1282 Forgery Attacks on GCM in TLS", May 2016, 1283 . 1285 [CAB-Baseline] 1286 CA/Browser Forum, "Baseline Requirements for the Issuance 1287 and Management of Publicly-Trusted Certificates Version 1288 1.1.6", 2013, . 1290 [Chung18] Chung, T., Lok, J., Chandrasekaran, B., Choffnes, D., 1291 Levin, D., Maggs, B., Mislove, A., Rula, J., Sullivan, N., 1292 and C. Wilson, "Is the Web Ready for OCSP Must-Staple?", 1293 Proceedings of the Internet Measurement Conference 2018, 1294 DOI 10.1145/3278532.3278543, October 2018, 1295 . 1297 [CRLite] Larisch, J., Choffnes, D., Levin, D., Maggs, B., Mislove, 1298 A., and C. Wilson, "CRLite: A Scalable System for Pushing 1299 All TLS Revocations to All Browsers", 2017 IEEE Symposium 1300 on Security and Privacy (SP), DOI 10.1109/sp.2017.17, May 1301 2017, . 1303 [CVE] MITRE, "Common Vulnerabilities and Exposures", 1304 . 1306 [DANE-SMTP] 1307 Dukhovni, V. and W. Hardaker, "SMTP Security via 1308 Opportunistic DNS-Based Authentication of Named Entities 1309 (DANE) Transport Layer Security (TLS)", RFC 7672, 1310 DOI 10.17487/RFC7672, October 2015, 1311 . 1313 [DANE-SRV] Finch, T., Miller, M., and P. Saint-Andre, "Using DNS- 1314 Based Authentication of Named Entities (DANE) TLSA Records 1315 with SRV Records", RFC 7673, DOI 10.17487/RFC7673, October 1316 2015, . 1318 [DegabrieleP07] 1319 Degabriele, J. and K. Paterson, "Attacking the IPsec 1320 Standards in Encryption-only Configurations", 2007 IEEE 1321 Symposium on Security and Privacy (SP '07), 1322 DOI 10.1109/sp.2007.8, May 2007, 1323 . 1325 [DEP-SSLv3] 1326 Barnes, R., Thomson, M., Pironti, A., and A. Langley, 1327 "Deprecating Secure Sockets Layer Version 3.0", RFC 7568, 1328 DOI 10.17487/RFC7568, June 2015, 1329 . 1331 [Heninger2012] 1332 Heninger, N., Durumeric, Z., Wustrow, E., and J.A. 1333 Halderman, "Mining Your Ps and Qs: Detection of Widespread 1334 Weak Keys in Network Devices", Usenix Security 1335 Symposium 2012, 2012. 1337 [I-D.ietf-tls-esni] 1338 Rescorla, E., Oku, K., Sullivan, N., and C. A. Wood, "TLS 1339 Encrypted Client Hello", Work in Progress, Internet-Draft, 1340 draft-ietf-tls-esni-13, 12 August 2021, 1341 . 1344 [I-D.irtf-cfrg-aead-limits] 1345 Günther, F., Thomson, M., and C. A. Wood, "Usage Limits on 1346 AEAD Algorithms", Work in Progress, Internet-Draft, draft- 1347 irtf-cfrg-aead-limits-03, 12 July 2021, 1348 . 1351 [IANA_TLS] IANA, "Transport Layer Security (TLS) Parameters", 1352 . 1354 [Jager2015] 1355 Jager, T., Schwenk, J., and J. Somorovsky, "Practical 1356 Invalid Curve Attacks on TLS-ECDH", European Symposium on 1357 Research in Computer Security (ESORICS) 2015 , 2015. 1359 [Joux2006] Joux, A., "Authentication Failures in NIST version of 1360 GCM", 2006, . 1364 [Kleinjung2010] 1365 Kleinjung, T., Aoki, K., Franke, J., Lenstra, A., Thomé, 1366 E., Bos, J., Gaudry, P., Kruppa, A., Montgomery, P., 1367 Osvik, D., te Riele, H., Timofeev, A., and P. Zimmermann, 1368 "Factorization of a 768-Bit RSA Modulus", Advances in 1369 Cryptology - CRYPTO 2010 pp. 333-350, 1370 DOI 10.1007/978-3-642-14623-7_18, 2010, 1371 . 1373 [LetsRevoke] 1374 Smith, T., Dickinson, L., and K. Seamons, "Let's Revoke: 1375 Scalable Global Certificate Revocation", Proceedings 2020 1376 Network and Distributed System Security Symposium, 1377 DOI 10.14722/ndss.2020.24084, 2020, 1378 . 1380 [Logjam] Adrian, D., Bhargavan, K., Durumeric, Z., Gaudry, P., 1381 Green, M., Halderman, J., Heninger, N., Springall, D., 1382 Thomé, E., Valenta, L., VanderSloot, B., Wustrow, E., 1383 Zanella-Béguelin, S., and P. Zimmermann, "Imperfect 1384 Forward Secrecy: How Diffie-Hellman Fails in Practice", 1385 Proceedings of the 22nd ACM SIGSAC Conference on Computer 1386 and Communications Security, DOI 10.1145/2810103.2813707, 1387 October 2015, . 1389 [Multiple-Encryption] 1390 Merkle, R. and M. Hellman, "On the security of multiple 1391 encryption", Communications of the ACM Vol. 24, pp. 1392 465-467, DOI 10.1145/358699.358718, July 1981, 1393 . 1395 [NIST.SP.800-56A] 1396 Barker, E., Chen, L., Roginsky, A., Vassilev, A., and R. 1397 Davis, "Recommendation for pair-wise key-establishment 1398 schemes using discrete logarithm cryptography", National 1399 Institute of Standards and Technology report, 1400 DOI 10.6028/nist.sp.800-56ar3, April 2018, 1401 . 1403 [PatersonRS11] 1404 Paterson, K., Ristenpart, T., and T. Shrimpton, "Tag Size 1405 Does Matter: Attacks and Proofs for the TLS Record 1406 Protocol", Lecture Notes in Computer Science pp. 372-389, 1407 DOI 10.1007/978-3-642-25385-0_20, 2011, 1408 . 1410 [POODLE] US-CERT, "SSL 3.0 Protocol Vulnerability and POODLE 1411 Attack", October 2014, 1412 . 1414 [RACCOON] Merget, R., Brinkmann, M., Aviram, N., Somorovsky, J., 1415 Mittmann, J., and J. Schwenk, "Raccoon Attack: Finding and 1416 Exploiting Most-Significant-Bit-Oracles in TLS-DH(E)", 1417 30th USENIX Security Symposium (USENIX Security 21) , 1418 2021, . 1421 [RFC2026] Bradner, S., "The Internet Standards Process -- Revision 1422 3", BCP 9, RFC 2026, DOI 10.17487/RFC2026, October 1996, 1423 . 1425 [RFC2246] Dierks, T. and C. Allen, "The TLS Protocol Version 1.0", 1426 RFC 2246, DOI 10.17487/RFC2246, January 1999, 1427 . 1429 [RFC3602] Frankel, S., Glenn, R., and S. Kelly, "The AES-CBC Cipher 1430 Algorithm and Its Use with IPsec", RFC 3602, 1431 DOI 10.17487/RFC3602, September 2003, 1432 . 1434 [RFC4346] Dierks, T. and E. Rescorla, "The Transport Layer Security 1435 (TLS) Protocol Version 1.1", RFC 4346, 1436 DOI 10.17487/RFC4346, April 2006, 1437 . 1439 [RFC4347] Rescorla, E. and N. Modadugu, "Datagram Transport Layer 1440 Security", RFC 4347, DOI 10.17487/RFC4347, April 2006, 1441 . 1443 [RFC5077] Salowey, J., Zhou, H., Eronen, P., and H. Tschofenig, 1444 "Transport Layer Security (TLS) Session Resumption without 1445 Server-Side State", RFC 5077, DOI 10.17487/RFC5077, 1446 January 2008, . 1448 [RFC5116] McGrew, D., "An Interface and Algorithms for Authenticated 1449 Encryption", RFC 5116, DOI 10.17487/RFC5116, January 2008, 1450 . 1452 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 1453 Housley, R., and W. Polk, "Internet X.509 Public Key 1454 Infrastructure Certificate and Certificate Revocation List 1455 (CRL) Profile", RFC 5280, DOI 10.17487/RFC5280, May 2008, 1456 . 1458 [RFC6101] Freier, A., Karlton, P., and P. Kocher, "The Secure 1459 Sockets Layer (SSL) Protocol Version 3.0", RFC 6101, 1460 DOI 10.17487/RFC6101, August 2011, 1461 . 1463 [RFC6120] Saint-Andre, P., "Extensible Messaging and Presence 1464 Protocol (XMPP): Core", RFC 6120, DOI 10.17487/RFC6120, 1465 March 2011, . 1467 [RFC6698] Hoffman, P. and J. Schlyter, "The DNS-Based Authentication 1468 of Named Entities (DANE) Transport Layer Security (TLS) 1469 Protocol: TLSA", RFC 6698, DOI 10.17487/RFC6698, August 1470 2012, . 1472 [RFC6797] Hodges, J., Jackson, C., and A. Barth, "HTTP Strict 1473 Transport Security (HSTS)", RFC 6797, 1474 DOI 10.17487/RFC6797, November 2012, 1475 . 1477 [RFC6960] Santesson, S., Myers, M., Ankney, R., Malpani, A., 1478 Galperin, S., and C. Adams, "X.509 Internet Public Key 1479 Infrastructure Online Certificate Status Protocol - OCSP", 1480 RFC 6960, DOI 10.17487/RFC6960, June 2013, 1481 . 1483 [RFC6961] Pettersen, Y., "The Transport Layer Security (TLS) 1484 Multiple Certificate Status Request Extension", RFC 6961, 1485 DOI 10.17487/RFC6961, June 2013, 1486 . 1488 [RFC7435] Dukhovni, V., "Opportunistic Security: Some Protection 1489 Most of the Time", RFC 7435, DOI 10.17487/RFC7435, 1490 December 2014, . 1492 [RFC7457] Sheffer, Y., Holz, R., and P. Saint-Andre, "Summarizing 1493 Known Attacks on Transport Layer Security (TLS) and 1494 Datagram TLS (DTLS)", RFC 7457, DOI 10.17487/RFC7457, 1495 February 2015, . 1497 [RFC7507] Moeller, B. and A. Langley, "TLS Fallback Signaling Cipher 1498 Suite Value (SCSV) for Preventing Protocol Downgrade 1499 Attacks", RFC 7507, DOI 10.17487/RFC7507, April 2015, 1500 . 1502 [RFC7525] Sheffer, Y., Holz, R., and P. Saint-Andre, 1503 "Recommendations for Secure Use of Transport Layer 1504 Security (TLS) and Datagram Transport Layer Security 1505 (DTLS)", BCP 195, RFC 7525, DOI 10.17487/RFC7525, May 1506 2015, . 1508 [RFC7590] Saint-Andre, P. and T. Alkemade, "Use of Transport Layer 1509 Security (TLS) in the Extensible Messaging and Presence 1510 Protocol (XMPP)", RFC 7590, DOI 10.17487/RFC7590, June 1511 2015, . 1513 [RFC7633] Hallam-Baker, P., "X.509v3 Transport Layer Security (TLS) 1514 Feature Extension", RFC 7633, DOI 10.17487/RFC7633, 1515 October 2015, . 1517 [RFC7919] Gillmor, D., "Negotiated Finite Field Diffie-Hellman 1518 Ephemeral Parameters for Transport Layer Security (TLS)", 1519 RFC 7919, DOI 10.17487/RFC7919, August 2016, 1520 . 1522 [RFC8452] Gueron, S., Langley, A., and Y. Lindell, "AES-GCM-SIV: 1523 Nonce Misuse-Resistant Authenticated Encryption", 1524 RFC 8452, DOI 10.17487/RFC8452, April 2019, 1525 . 1527 [RFC8470] Thomson, M., Nottingham, M., and W. Tarreau, "Using Early 1528 Data in HTTP", RFC 8470, DOI 10.17487/RFC8470, September 1529 2018, . 1531 [RFC9001] Thomson, M., Ed. and S. Turner, Ed., "Using TLS to Secure 1532 QUIC", RFC 9001, DOI 10.17487/RFC9001, May 2021, 1533 . 1535 [RFC9162] Laurie, B., Messeri, E., and R. Stradling, "Certificate 1536 Transparency Version 2.0", RFC 9162, DOI 10.17487/RFC9162, 1537 December 2021, . 1539 [SAFECURVES] 1540 Bernstein, D.J. and T. Lange, "SafeCurves: Choosing Safe 1541 Curves for Elliptic-Curve Cryptography", December 2014, 1542 . 1544 [Soghoian2011] 1545 Soghoian, C. and S. Stamm, "Certified Lies: Detecting and 1546 Defeating Government Interception Attacks Against SSL", 1547 SSRN Electronic Journal, DOI 10.2139/ssrn.1591033, 2010, 1548 . 1550 [Springall16] 1551 Springall, D., Durumeric, Z., and J. Halderman, "Measuring 1552 the Security Harm of TLS Crypto Shortcuts", Proceedings of 1553 the 2016 Internet Measurement Conference, 1554 DOI 10.1145/2987443.2987480, November 2016, 1555 . 1557 [Sy2018] Sy, E., Burkert, C., Federrath, H., and M. Fischer, 1558 "Tracking Users across the Web via TLS Session 1559 Resumption", Proceedings of the 34th Annual Computer 1560 Security Applications Conference, 1561 DOI 10.1145/3274694.3274708, December 2018, 1562 . 1564 [triple-handshake] 1565 Bhargavan, K., Lavaud, A., Fournet, C., Pironti, A., and 1566 P. Strub, "Triple Handshakes and Cookie Cutters: Breaking 1567 and Fixing Authentication over TLS", 2014 IEEE Symposium 1568 on Security and Privacy, DOI 10.1109/sp.2014.14, May 2014, 1569 . 1571 Appendix A. Differences from RFC 7525 1573 This revision of the Best Current Practices contains numerous 1574 changes, and this section is focused on the normative changes. 1576 * High level differences: 1578 - Clarified items (e.g. renegotiation) that only apply to TLS 1579 1.2. 1581 - Changed status of TLS 1.0 and 1.1 from SHOULD NOT to MUST NOT. 1583 - Added TLS 1.3 at a SHOULD level. 1585 - Similar changes to DTLS, pending publication of DTLS 1.3. 1587 - Specific guidance for multiplexed protocols. 1589 - MUST-level implementation requirement for ALPN, and more 1590 specific SHOULD-level guidance for ALPN and SNI. 1592 - Limits on key usage. 1594 - New attacks since [RFC7457]: ALPACA, Raccoon, Logjam, "Nonce- 1595 Disrespecting Adversaries". 1597 - RFC 6961 (OCSP status_request_v2) has been deprecated. 1599 * Differences specific to TLS 1.2: 1601 - SHOULD-level guidance on AES-GCM nonce generation. 1603 - SHOULD NOT use (static or ephemeral) finite-field DH key 1604 agreement. 1606 - SHOULD NOT reuse ephemeral finite-field DH keys across multiple 1607 connections. 1609 - 2048-bit DH now a MUST, ECDH minimal curve size is 224, vs. 192 1610 previously. 1612 - Support for extended_master_secret is a SHOULD. Also removed 1613 other, more complicated, related mitigations. 1615 - SHOULD-level restriction on the TLS session duration, depending 1616 on the rotation period of an [RFC5077] ticket key. 1618 - Drop TLS_DHE_RSA_WITH_AES from the recommended ciphers 1620 - Add TLS_ECDHE_ECDSA_WITH_AES to the recommended ciphers 1622 - SHOULD NOT use the old MTI cipher suite, 1623 TLS_RSA_WITH_AES_128_CBC_SHA. 1625 - Recommend curve X25519 alongside NIST P-256 1627 * Differences specific to TLS 1.3: 1629 - New TLS 1.3 capabilities: 0-RTT. 1631 - Removed capabilities: renegotiation, compression. 1633 - Added mention of TLS Encrypted Client Hello, but no 1634 recommendation to use until it is finalized. 1636 - SHOULD-level requirement for forward secrecy in TLS 1.3 session 1637 resumption. 1639 - Generic SHOULD-level guidance to avoid 0-RTT unless it is 1640 documented for the particular protocol. 1642 Appendix B. Document History 1644 // Note to RFC Editor: please remove before publication. 1646 B.1. draft-ietf-uta-rfc7525bis-05 1648 * Addressed WG Last Call comments, specifically: 1650 - More clarity and guidance on session resumption. 1652 - Clarity on TLS 1.2 renegotiation. 1654 - Wording on the 0-RTT feature aligned with RFC 8446. 1656 - SHOULD NOT guidance on static and ephemeral finite field DH 1657 cipher suites. 1659 - Revamped the recommended TLS 1.2 cipher suites, removing DHE 1660 and adding ECDSA. The latter due to the wide adoption of ECDSA 1661 certificates and in line with RFC 8446. 1663 - Recommendation to use deterministic ECDSA. 1665 - Finally deprecated the old TLS 1.2 MTI cipher suite. 1667 - Deeper discussion of ECDH public key reuse issues, and as a 1668 result, recommended support of X25519. 1670 - Reworded the section on certificate revocation and OCSP 1671 following a long mailing list thread. 1673 B.2. draft-ietf-uta-rfc7525bis-04 1675 * No version fallback from TLS 1.2 to earlier versions, therefore no 1676 SCSV. 1678 B.3. draft-ietf-uta-rfc7525bis-03 1680 * Cipher integrity and confidentiality limits. 1682 * Require extended_master_secret. 1684 B.4. draft-ietf-uta-rfc7525bis-02 1686 * Adjusted text about ALPN support in application protocols 1688 * Incorporated text from draft-ietf-tls-md5-sha1-deprecate 1690 B.5. draft-ietf-uta-rfc7525bis-01 1692 * Many more changes, including: 1694 - SHOULD-level requirement for forward secrecy in TLS 1.3 session 1695 resumption. 1697 - Removed TLS 1.2 capabilities: renegotiation, compression. 1699 - Specific guidance for multiplexed protocols. 1701 - MUST-level implementation requirement for ALPN, and more 1702 specific SHOULD-level guidance for ALPN and SNI. 1704 - Generic SHOULD-level guidance to avoid 0-RTT unless it is 1705 documented for the particular protocol. 1707 - SHOULD-level guidance on AES-GCM nonce generation in TLS 1.2. 1709 - SHOULD NOT use static DH keys or reuse ephemeral DH keys across 1710 multiple connections. 1712 - 2048-bit DH now a MUST, ECDH minimal curve size is 224, up from 1713 192. 1715 B.6. draft-ietf-uta-rfc7525bis-00 1717 * Renamed: WG document. 1719 * Started populating list of changes from RFC 7525. 1721 * General rewording of abstract and intro for revised version. 1723 * Protocol versions, fallback. 1725 * Reference to ECHO. 1727 B.7. draft-sheffer-uta-rfc7525bis-00 1729 * Renamed, since the BCP number does not change. 1731 * Added an empty "Differences from RFC 7525" section. 1733 B.8. draft-sheffer-uta-bcp195bis-00 1735 * Initial release, the RFC 7525 text as-is, with some minor 1736 editorial changes to the references. 1738 Authors' Addresses 1739 Yaron Sheffer 1740 Intuit 1742 Email: yaronf.ietf@gmail.com 1744 Peter Saint-Andre 1745 Mozilla 1747 Email: stpeter@mozilla.com 1749 Thomas Fossati 1750 arm 1752 Email: thomas.fossati@arm.com